Trans-regulation of Syndecan Functions by Hetero-oligomerization*

Background: Syndecans form non-covalently linked homodimers through their highly conserved transmembrane domains. Results: Syndecan-2 and -4 exhibit heteromolecular interaction, and this interaction regulates syndecan-mediated cellular functions. Conclusion: Hetero-oligomerization produces distinct syndecan functions. Significance: Our finding offers new insights into the underlying signaling mechanisms of syndecans. Syndecans, a family of transmembrane heparansulfate proteoglycans, are known to interact through their transmembrane domains to form non-covalently linked homodimers, a process essential for their individual functions. Because all syndecan transmembrane domains are highly conserved and thus might mediate interactions between different members of the syndecan family, we investigated syndecan interactions in detail. All recombinant syndecan-2 and -4 protein variants containing the transmembrane domain formed not only sodium dodecyl sulfate (SDS)-resistant homodimers but also SDS-resistant heterodimers. Biochemical and structural data revealed that recombinant syndecan-2 and -4 formed intermolecular interactions in vitro, and the GXXXG motif in transmembrane domain mediated this interaction. When exogenously expressed in rat embryonic fibroblasts, syndecan-2 interacted with syndecan-4 and vice versa. Furthermore, bimolecular fluorescence complementation-based assay demonstrated specific hetero-molecular interactions between syndecan-2 and -4, supporting hetero-oligomer formation of syndecans in vivo. Interestingly, hetero-oligomerization significantly reduced syndecan-4-mediated cellular processes such as protein kinase Cα activation and protein kinase Cα-mediated cell adhesion as well as syndecan-2-mediated tumorigenic activities in colon cancer cells such as migration and anchorage-independent growth. Taken together, these data provide evidence that hetero-oligomerization produces distinct syndecan functions and offer insights into the underlying signaling mechanisms of syndecans.


Syndecans, a family of transmembrane heparansulfate proteoglycans, are known to interact through their transmembrane domains to form non-covalently linked homodimers, a process essential for their individual functions. Because all syndecan transmembrane domains are highly conserved and thus might mediate interactions between different members of the syndecan family, we investigated syndecan interactions in detail. All recombinant syndecan-2 and -4 protein variants containing the transmembrane domain formed not only sodium dodecyl sulfate (SDS)-resistant homodimers but also SDS-resistant heterodimers. Biochemical and structural data revealed that recombinant syndecan-2 and -4 formed intermolecular interactions in vitro
, and the GXXXG motif in transmembrane domain mediated this interaction. When exogenously expressed in rat embryonic fibroblasts, syndecan-2 interacted with syndecan-4 and vice versa. Furthermore, bimolecular fluorescence complementation-based assay demonstrated specific hetero-molecular interactions between syndecan-2 and -4, supporting hetero-oligomer formation of syndecans in vivo. Interestingly, hetero-oligomerization significantly reduced syndecan-4-mediated cellular processes such as protein kinase C␣ activation and protein kinase C␣-mediated cell adhesion as well as syndecan-2-mediated tumorigenic activities in colon cancer cells such as migration and anchorage-independent growth. Taken together, these data provide evidence that hetero-oligomerization produces distinct syndecan functions and offer insights into the underlying signaling mechanisms of syndecans.
Syndecans are a family of evolutionarily conserved transmembrane heparan sulfate proteoglycans that participate in the regulation of a broad range of adhesion-mediated cell functions (1). Structurally, they are composed of an extracellular domain, a single transmembrane domain, and a short cytoplasmic domain. The extracellular domain of syndecans interacts with a number of ligands in the extracellular matrix through heparan sulfate chains and interactions that activate syndecans and lead to their regulation of receptor signaling events (1)(2)(3). The transmembrane domain is composed of 25 hydrophobic amino acid residues that are responsible for the molecular interaction that causes homo-oligomerization of syndecan core proteins (4,5). The strictly conserved GXXXG motif is involved in these interactions and induces homodimer formation of syndecans, even in the presence of the strong anionic detergent sodium dodecyl sulfate (SDS), a property known as SDS-resistant dimerization (4). The cytoplasmic domain contains two highly conserved C1 and C2 regions separated by a variable region. The intervening variable region, which is distinct for each syndecan member yet conserved across species, promotes syndecan-specific intracellular functions.
Four members of the syndecan family have been identified in vertebrates (6), and their expression is highly regulated in complex patterns in individual cell types and tissues and at different developmental stages (7). Syndecan-1 is the predominant family member in epithelial cells, syndecan-2 is abundantly expressed in fibroblasts and mesenchymal cells, syndecan-3 abounds in neuronal cells, and syndecan-4 is widely expressed (7). Notably, the expression of syndecans is altered in pathological situations such as wound healing and neoplastic transformation. In particular, altered expression of syndecan family members has been reported in various human tumors, including lung, breast, and colorectal cancers (8,9).
Transduction of signals through non-covalent dimerization/ oligomerization of cell surface receptor proteins containing a single transmembrane domain in response to ligand binding is a well-recognized signaling paradigm (10). As is the case for many other cell surface receptors, homodimerization/oligomerization is the first step in the activation of syndecan signaling, and the syndecan transmembrane domain plays a major role in this process (5). Given the high pairwise sequence identity of syndecan transmembrane domains (48 -72%) and the conservation of the GXXXG dimerization motif, all syndecan transmembrane domains might be expected to mediate heteromeric oligomerization as well as homomeric interactions. A previous study using synthetic peptides corresponding to the transmem-brane domain of each syndecan paralog has measured the association tendencies of the syndecan transmembrane domain (11). This study found that the transmembrane domains of the four syndecan paralogs self-associate to very different degrees and are capable of a distinct set of heteromeric interactions, independent of extracellular and cytoplasmic domains (11). However, whether full-length syndecans form heterodimers and, if so, how this heterodimerization is involved in regulating syndecan functions, have not been investigated. Here, we demonstrate that transmembrane domain-mediated hetero-oligomerization of syndecan-2 with syndecan-4 occurs in vitro and in intact cells and regulates syndecan-mediated signal transduction, producing unique functional outcomes.

Experimental Procedures
Antibodies and Materials-Monoclonal antibodies against glutathione S-transferase (GST), His, HA, ␤-actin, RhoA, integrin ␤1, extracellular signal-regulated kinase (ERK) and phospho-ERK, and polyclonal antibodies against Myc and protein kinase C␣ (PKC␣) were purchased from Santa Cruz Biotechnology (Santa Cruz, CA). A monoclonal antibody to ␣-actinin was purchased from Sigma, and a polyclonal antibody to syndecan-4 was purchased from Santa Cruz Biotechnology. Monoclonal antibodies to paxillin and Rac1 were purchased from Upstate Biotechnology, Inc. (Lake Placid, NY), and a monoclonal anti-syndecan-2 antibody was produced by AdipoGen Inc. (Incheon, Korea). Fibronectin was from Upstate Biotechnology, Inc. (Lake Placid, NY).Transient transfection was carried out using Effectene (Qiagen, Hilden, Germany) as described by the provided protocol.
Expression and Purification of Recombinant GST-Syndecan Core Proteins-cDNAs of rat syndecan-1, -2, and -4 encoding full-length proteins (SDC1, SDC2, SDC4), the transmembrane domain and cytoplasmic domain (2TC, 4TC), the extracellular domain (2E), the transmembrane domain (2T), the cytoplasmic domain (2C), or the transmembrane domain containing four additional flanking extracellular amino acid residues (KRTE and ERTE) in the extracellular domain and the cytoplasmic domain (2eTC, 4eTC) were synthesized by polymerase chain reaction (PCR) and subcloned into the GST expression vector pGEX-5X-1 (Amersham Biosciences). Mutations were generated by PCR-based site-directed mutagenesis (Stratagene, La Jolla, CA) and confirmed by DNA sequence analysis of the resulting plasmids. These constructs were used to transform Escherichia coli DH5␣, and the expression of GST fusion proteins was induced by incubation with 1 mM isopropyl-␤-D-thiogalactopyranoside for 4 h at 37°C. The fusion proteins were purified with glutathione-agarose beads (GE Healthcare) as described previously (5).
Purification of Recombinant His-syndecan Proteins by GST-Syndecan-bound Glutathione-agarose Bead Affinity Chromatography-cDNA encoding the entire rat syndecan-2 or -4 core protein was subcloned into the His expression vector pET32aϩ (Novagen, Madison, WI), and the expression of fusion proteins in E. coli BL21 was induced by incubating with 0.3 mM isopropyl-␤-D-thiogalactopyranoside at 25°C for 16 h. Proteins were released by lysing E. coli cells with lysis buffer (20 mM Na 2 HPO 4 , pH 8.0, 150 mM NaCl, 5 mM ␤-mercaptoethanol, 1% Triton X-100) and sonicating on ice for 1 m. After removing insoluble material by centrifugation at 13,000 ϫ g for 30 min at 4°C, the supernatants containing His-syndecan fusion proteins were applied to a glutathione-agarose column containing prebound GST-syndecans. The column was washed 3 times, and bound proteins were eluted with elution buffer (50 mM Tris-HCl, pH 8.0, 5 mM reduced glutathione). Fractions were analyzed by SDS-PAGE followed by Coomassie Blue staining and Western blotting using antibodies against GST, His, syndecan-2, and syndecan-4.
Sample Preparation for the Nuclear Magnetic Resonance (NMR) Experiment-cDNA encoding the rat transmembrane domain of syndecan-2 and syndecan-4 were subcloned into the His-thioredoxin expression vector pET32aϩ, and the enterokinase enzyme recognition site, DDDDK, was inserted between His-thioredoxin tag and target protein. Fusion protein expression was induced in E. coli BL21(DE3) cells with 1 mM isopropyl-␤-D-thiogalactopyranoside in optical density values of 0.55 at 600 nm and overexpressed at 25°C for 18 h. Harvested cell pellet was lysed with lysis buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM ␤-mercaptoethanol) and sonicated on ice for 1 m. After centrifugation at 13,000 ϫ g for 30 min, supernatant was removed, and insoluble precipitant was used for the resolubilization step using refolding buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM ␤-mercaptoethanol, 1% n-dodecyl ␤-Dmaltoside (Affymetrix). Resolubilized fraction was applied to Ni-NTA affinity column. The column was washed two times using 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM ␤-mercaptoethanol, 20 mM imidazole, and 0.1% n-dodecyl ␤-D-maltoside, and target protein was eluted in 50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 5 mM ␤-mercaptoethanol, 250 mM imidazole, 0.05% n-dodecyl ␤-D-maltoside. After desalting column work using PD-10 (GE Healthcare), enterokinase enzyme was treated to remove the tag at 37°C for 12 h. Nonspecific impurity and His-thioredoxin tag was removed using additional Ni-NTA affinity column work. Final target protein was dialyzed in 20 mM ammonium bicarbonate and lyophilized. For the preparation of NMR samples, M9 minimal media and isotopes such as [ 13 C]glucose and 15 NH 4 Cl were used during cell culture. All NMR samples are prepared in 10 mM Na 2 HPO 4 , pH 7.0, 4 mM DTT, 1 mM sodium azide, 100 mM dodecylphosphocholine, 10% D 2 O.
Immunoprecipitation-For the immunoprecipitation of HAtagged syndecans, REFs transfected with the indicated cDNAs were lysed with 1% Nonidet P-40 lysis buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 1 mM EDTA, 2 mM Na 3 VO 4 , and 10 mM NaF) containing a protease inhibitor mixture. Individual cell lysates were mixed for 30 min on ice, and protein mixtures were incubated with anti-HA affinity Matrix (Roche Applied Science) for 2 h at 4°C. Immune complexes were collected by centrifugation, washed 3 times with lysis buffer, eluted with 1 M NaCl, and analyzed by slot blotting using the indicated antibodies. For the immunoprecipitation of endogenous syndecan proteins, cells were lysed in 1% Nonidet P-40 lysis buffer containing a protease inhibitor mixture. Each cell lysate was incubated with anti-syndecan antibodies for 2 h at 4°C followed by incubation with protein G-Sepharose beads for 1 h. Immune complexes were collected by centrifugation, washed 3 times with lysis buffer, and eluted with 1 M NaCl. To remove the IgG of eluted proteins and extract the syndecan antibodies, eluted proteins were incubated with DEAE-Sepharose beads (GE Healthcare) for 2 h at 4°C in 50 mM Tris, pH 8.0, containing 4 M urea, 200 mM NaCl, 0.1% Tween 20, 5 mM EDTA, and 2 mM PMSF, washed with 30 mM sodium acetate, pH 4.0, containing 4 M urea, 200 mM NaCl, 0.1% Tween 20, 25 mM EDTA, and 2 mM PMSF, eluted with 15 mM sodium acetate buffer, pH 4.0, containing 4 M guanidine HCl and 0.1% Tween-20, and then analyzed by slot blotting using the indicated antibodies. For Western blotting of immunoprecipitates, cells were lysed with 1% radioimmune precipitation assay lysis buffer (50 mM Tris, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.1% SDS, 0.5% sodium deoxycholate, 10 mM NaF, 2 mM Na 3 VO 4 ) containing a protease inhibitor mixture. Cell lysates were incubated with anti-syndecan-4 antibody for 2 h at 4°C followed by incubation with protein A-Sepharose beads for 1 h. Immune complexes were collected by centrifugation, washed 3 times with lysis buffer, and eluted with 1 M NaCl. Glycosaminoglycan chains were removed by incubating samples with a pH 1.5 nitrous acid solution as previously described (12) and analyzed by Western blotting using an anti-syndecan-2 antibody.
BimolecularFluorescenceComplementation-Thevectorconstructs encoding fragments of pECFP were kindly provided by Dr. Minsoo Kim (University of Rochester Medical Center). In the current study the sequences encoding pECFP amino acid residues 1-154 (CN) or 155-238 (CC) were fused to the 3Ј ends of the coding regions of syndecans by conventional restriction enzyme-based cloning; the resulting constructs generated are depicted schematically in Fig. 4E. For bimolecular fluorescence complementation assays, HT29 cells were transiently co-transfected with the indicated cDNAs then fixed for 5 min in 4% paraformaldehyde, washed with PBS, and imaged on a Zeiss Axiovert 200 M inverted microscope equipped with a Zeiss 510 META confocal head using a C-Apochromat 40ϫ/1.2-W objective lens and a 458-nm argon laser light and Meta detector (462.6 -516.1 nm) for pECFP excitation and emission.
In Vitro PKC Assay-For nonradioactive Peptag assays, Pep-Tag C1 peptide (Promega, Madison, WI) was incubated in reaction buffer (20 mM HEPES, pH 7.4, 1 mM DTT, 10 mM MgCl 2 , 1 mM ATP) containing phosphatidylinositol 4,5-bisphosphate (50 M) in a final volume of 25 l for 30 min at 30°C. The reactions were stopped by heating to 95°C for 10 m. The samples were separated on a 0.8% agarose gel at 100 V for 15 min. Phosphorylated peptides migrated toward the cathode (ϩ), whereas non-phosphorylated peptides migrated toward the anode (Ϫ). The gel was photographed on a transilluminator.
Centrifugal Detachment Assay-Cell-substratum adhesiveness was quantified using an inverted centrifugal detachment assay. Fibronectin was diluted in serum-free medium (10 g/ml), added to the plates, and incubated at 25°C for 1 h. The plates were then washed with PBS and blocked with 0.2% heat-inactivated BSA for 1 h. After washing with PBS, cells were incubated for additional 2 h at 37°C in 5% CO 2 . After removing unattached cells, plates were then filled with serum-free medium, sealed with Parafilm, and centrifuged inverted for 5 min at 300 ϫ g at room temperature using a large capacity tabletop centrifuge (Hanil science industrial, Korea). After the centrifugation, retained cells were counted using a hemocytometer.
Cellular Fractionation-After washing twice with PBS, a hypo-osmotic solution (20 mM Tris/HCl, pH 7.5, 2 mM 2-mercaptoethanol, 5 mM EGTA, 2 mM EDTA) containing a protease inhibitor mixture was added to the culture plates. Cells were subsequently scraped off the plates and homogenized on ice. The homogenate was centrifuged at 13,000 ϫ g for 15 min at 4°C. The membrane fraction was collected by solubilizing the remaining pellet in radioimmune precipitation assay buffer containing a protease inhibitor mixture and then centrifuged at 13,000 ϫ g for 15 min at 4°C. Equal amounts of lysates were resolved by SDS-PAGE, transferred onto PVDF membranes, and probed with the indicated antibodies.
Rac and Rho Activity Assay-GST-PAK-PBD binding assays were performed essentially as described previously (13). Briefly, the p21 binding domain of PAK1 (PBD) was expressed in E. coli as a GST-PAK-PBD fusion protein, purified using glutathioneagarose beads, and added to cell lysates. Bound proteins were collected by centrifugation and suspended in SDS sample buffer. Proteins were fractionated by SDS-PAGE and transferred to PVDF membranes, and the amount of precipitated Rac1 was estimated by Western blotting with an anti-Rac1 antibody. Rho activity was measured in a pulldown assay using the Rho binding domain from Rhotekin. Equal volumes of lysates were incubated with GST-Rho binding domain beads at 4°C for 2 h, after which the beads were washed 4 times with lysis buffer, and bound RhoA proteins were detected by Western blotting using a monoclonal antibody against RhoA.
Immunofluorescence-Cells were transfected with the indicated cDNAs, fixed with 3.5% paraformaldehyde in PBS at room temperature for 5 min, permeabilized with 0.1% Triton X-100 in PBS for 10 min, blocked with 0.5% bovine serum albu-men in PBS for 1 h, and incubated with the indicated antibodies. The slides were mounted with VECTASHIELD Mounting Medium (Vector Laboratories, Burlingame, CA) and imaged using a fluorescence confocal microscope (Carl Zeiss, Gottingen, Germany).
Monitoring Cell Spreading and Migration-Cell spreading and migration were monitored using the xCELLigence system (Roche Diagnostics). For cell spreading, E-plate 16 assemblies (Roche Diagnostics) were coated with fibronectin (10 g/ml) and seeded with cells (1.5 ϫ 10 4 cells/well). Each plate was then assembled on the RTCA DP Analyzer, and data were gathered at 5-min intervals for 5 h at 37°C, 5% CO 2 . The data were analyzed using the provided RTCA software. For cell migration, a CIM-plate 16 (Roche Diagnostics) was used. The lower chambers were filled with fresh medium (160 l/well) containing 10 g/ml fibroblast growth factor-2 and 10% FBS, and the upper chambers were filled with medium (30 l/well) containing 5% FBS. Plates were incubated at 37°C, 5% CO 2 for 1 h, and background was measured using the RTCA DP Analyzer. Transfected cells were added to each well, and the plate was incubated at 25°C. After 30 min, the CIM-plate was assembled onto the RTCA DP Analyzer, and cell migration was assessed at 15-min intervals for 24 h at 37°C, 5% CO 2 . The obtained data were analyzed using the provided RTCA software.
Transwell Migration Assay-The lower surface of Transwell inserts (Costar) was coated with gelatin (10 g/ml), and the membranes were allowed to dry for 1 h at room temperature. The Transwell inserts were assembled into a 24-well plate, and the lower chamber was filled with McCoy's 5a media containing 10% FBS and fibroblast growth factor-2 (10 g/ml). Cells (5 ϫ 10 5 ) were added to each upper chamber, and the plate was incubated at 37°C in a 5% CO 2 incubator for 30 h. Migrated cells were stained with 0.6% hematoxylin and 0.5% eosin and counted.
Anchorage-independent Growth in Soft Agarose-Each well of a 6-well culture plate was coated with 3 ml of bottom agar mixture (McCoy's 5a, 10% FBS, 0.6% agar). After the bottom layer had solidified, 2 ml of top agar mixture (McCoy's 5a, 10% fetal bovine serum, 0.3% agar) containing 2 ϫ 10 6 cells was added to each well, and the cultures were incubated at 37°C in a 5% CO 2 atmosphere. Every 5 days, normal growth medium was gently layered over the cultures. Colony formation was monitored daily with a light microscope. Colonies in soft agar were photographed with a digital camera after incubation for 21 days.
Statistical Analysis-Data are represented as the mean from at least three independent experiments. Statistical analysis was performed using an unpaired Student's t test. A p value less than 0.01 was considered statistically significant.

Syndecans Form SDS-resistant Heterodimers through Transmembrane Domain
Interactions-To examine interactions of syndecan molecules, we first investigated the well known property of SDS-resistant dimer formation. Several recombinant syndecan-2 and -4 constructs were generated and expressed in E. coli as GST-fusion proteins (Fig. 1A). Recombinant syndecan proteins were induced with isopropyl-␤-D-thiogalactopyrano-side and purified. wild type syndecans (SDC2 or SDC4) were mixed with deletion mutants of syndecan-2 or -4 containing transmembrane and cytoplasmic domains (2TC and 4TC, respectively). Consistent with a previous report (5), recombinant syndecan-2 and -4 proteins were resolved as SDS-resistant homomeric dimers and monomers (Fig. 1B, left). Interestingly, mixtures of syndecan-2 or -4 proteins with deletion mutants showed only one extra band between the monomer-and dimermigrating positions, corresponding to a heteromeric molecular mass of ϳ90 kDa (Fig. 1B, right; Tables 1 and 2). Consistently, His-tagged wild type syndecan proteins showed strong SDSresistant heterodimers (Fig. 1C). Furthermore, His-tagged syndecan-2 formed SDS-resistant dimer with the tag-free recombinant syndecan-4, and the tag-free recombinant syndecan-2 also formed SDS-resistant dimer with His-tagged syndecan-4 ( Fig. 1D). In addition, recombinant syndecan-1 formed SDSresistant heterodimers with syndecan-2 or -4 (data not shown). Collectively, these data indicate that syndecans form strong SDS-resistant heterodimers as well as homodimers in vitro.
Because the highly conserved transmembrane domain of syndecan core proteins is responsible for the formation of stable, non-covalent homodimeric complexes, we next investigated whether the transmembrane domain is necessary for heterodimerization. As expected, wild type syndecan-4 formed SDS-resistant heterodimers with the deletion mutant containing the transmembrane domain but not formed with those containing either the cytoplasmic domain (2C) or the extracellular domain (2E) alone (Fig. 1E). To further demonstrate the role of the transmembrane domain in heterodimerization, we used syndecan mutants defective in transmembrane domain-mediated homo-oligomerization (2EPT2C, 4EPT2C, 2GL, or 4GL; see Fig. 1A and Ref. 5). Oligomerization-defective mutations abolished SDS-resistant heterodimer formation by syndecan-2 or -4 (Fig. 1F, bottom), supporting the notion that the transmembrane domain is crucial for SDS-resistant heterodimerization of syndecans.
Transmembrane Domain Mediates the Intermolecular Interaction between Syndecans-To investigate hetero-molecular interactions of the syndecan transmembrane domain, we immobilized GST-syndecan fusion proteins on glutathione beads and used them to purify their respective binding partners among various His-tagged syndecans ( Fig. 2A). We found that His-tagged syndecan-2 could be purified from GST-syndecan-2 immobilized on glutathione beads (Fig. 2C), whereas GST protein alone could not (Fig. 2B), supporting the homomeric association of syndecan-2. Notably, His-syndecan-4 fusion protein could also be purified from glutathione beadimmobilized GST-syndecan-2, probably through transmembrane domain-mediated hetero-molecular interactions (Fig.  2D). In the reciprocal experiment, His-syndecan-2 fusion protein was purified from GST-syndecan-4 immobilized on glutathione beads (Fig. 2E). These data indicate that, in addition to well known intermolecular interactions involving syndecans of a single type, hetero-molecular interactions occur between different syndecan members.
We next carried out a set of NMR experiments to examine the intermolecular interaction between syndecan-2 and -4 transmembrane domains in solution. As shown in Fig. 3A, in the presence of syndecan-2 transmembrane domain, the chemical shift perturbation of glycine residues of syndecan-4 transmembrane domain were observed, indicating the participation of GXXXG motif in the intermolecular interaction. In addition, evident chemical shift perturbations were observed in Ile-155, Gly-157, Gly-158, Val-159, Val-160, Val-166, and Phe-167 residues in the transmembrane domain of syndecan-4 upon syn-decan-2 interaction (Fig. 3B). Particularly, the most dramatic perturbations were observed in the residues, Gly-157, Val-159, and Val-160 residues (Fig. 3C). The intensity ratio between syndecan-4 homodimer and heterodimer of syndecan-2 and -4 was 1:3 based on peak intensities of the heteronuclear single quantum correlation spectrum (Fig. 3D), implying that the molecular interaction of syndecan-4 transmembrane domain to synde-

TABLE 1
The extra band of GST-tagged recombinant protein mixture was identified to be heparan sulfate proteoglycan core protein (syndecan-2 and -4) by mass spectrometry LC-MS/MS sequencing results of 10 peptides, all of which are the matched peptides that led to sequence coverage of 44%, provided identification of syndecan-2 core protein (Rattus norvegicus). Nominal molecular weight (M r ) was 22,135.

Syndecan Hetero-oligomerization
can-2 is stronger than that of syndecan-4. Taken together, these data confirm that syndecan-2 and -4 form heteromeric interactions in vitro and that the transmembrane domain mediates the intermolecular interaction between syndecans. Syndecan-2 Interacts with Syndecan-4 -Because syndecans are modified with highly negatively charged glycosaminoglycan chains that may disrupt their interactions through charge repulsion, the hetero-oligomerization characteristics of endogenously expressed syndecans may differ from those of recombinant proteins. To investigate the intermolecular interactions of syndecans at the cellular level, we separately expressed HAor Myc-tagged syndecan-2 or -4 in rat embryonic fibroblasts, and equal amounts of total cell lysate from REF-HA-syndecan-2 or REF-HA-syndecan-4 cells were mixed with those from REF-Myc-syndecan-2 or REF-Myc-syndecan-4 cells. Each mixture of total cell lysates was then immunoprecipitated with an anti-HA antibody, and the immunoprecipitate was analyzed by slot blotting using an anti-Myc antibody. As shown in Fig. 4A, HA-tagged syndecan-2 was co-immunoprecipitated with Myctagged syndecan-2 (Fig. 4A, bottom, left). A similar interaction of HA-tagged syndecan-4 and Myc-tagged syndecan-4 was also observed (Fig. 4A, bottom, right), supporting homo-oligomerization of both syndecan-2 and  (Fig. 4B, bottom). In addition, both slot blotting (Fig. 4C) and Western blotting (Fig. 4D) with appropriate anti-syndecan antibodies showed co-immunoprecipitation of endogenous syndecan-2 and -4, suggesting that syndecan-2 and -4 form both homomeric and heteromeric interactions in intact cells.
To further demonstrate the interaction of syndecan-2 with syndecan-4 at the cellular level, we employed a bimolecular fluorescence complementation assay. This protein-protein interaction technique relies on the ability of molecular fragments of a fluorescent protein, in this case fragments of pECFP (enhanced cyan fluorescent protein) fused to syndecan-2 and -4 ( Fig. 4E), to re-associate and refold into a fluorescent structure. Cells were co-transfected with CN and CC fusion constructs of pECFP with syndecan-2 (2CN and 2CC), syndecan-4 (4CN and 4CC), and subsequently analyzed by laser-scanning confocal microscopy for fluorescence complementation. Under each cotransfection condition, fluorescence complementation of both 2CC and 2CN and of 4CC and 4CN was detected at the cell surface (Fig. 4F). Similarly, fluorescence complementation of both 2CC and 4CN and of 4CC and 2CN was detected (Fig. 4F), indicating homomeric association of both syndecan-2 and syndecan-4 and heteromeric interactions of syndecan-2 and -4. However, we failed to detect complementation of fluorescence between wild type syndecan and oligomerization-defective mutants (Fig. 4F). Collectively, these data strongly suggest that, when expressed in the same cell, syndecan-2 and -4 are capable of both homo-oligomerization and hetero-oligomerization.

Hetero-oligomerization Inhibits Syndecan-regulated Molecular Events That Depend on Homo-oligomerization-Because
homo-oligomerization is crucial for syndecan functions, hetero-oligomer formation, which inevitably decreases homooligomerization, might result in inhibition of known syndecan functions. Previously, we and others have shown that oligomerization of the syndecan-4 cytoplasmic domain is critical for the interaction with and activation of PKC␣ (5,14,15). Solutions of purified wild type syndecan-2, purified 4eTC (a syndecan-4 deletion mutant containing the transmembrane domain with a 4-amino acid extracellular region and the cytoplasmic domain), or a mixture of 4eTC and wild type syndecan-2 were resolved by

FIGURE 2. Syndecans form both homomeric and heteromeric associations with partner syndecans.
A, scheme for purifying His-syndecan proteins using GST-syndecans immobilized on glutathione beads. #1, cell lysates from E. coli expressing GST-syndecan fusion proteins were applied to a column of glutathione-agarose beads, and beads were washed with lysis buffer containing 1% Triton X-100. #2, cell lysates from E. coli expressing His-syndecan fusion proteins were applied to a column of glutathione-agarose beads containing bound GST-syndecan fusion proteins, and beads were washed with His lysis buffer containing 1% Triton X-100. #3, bound materials were eluted with elution buffer containing reduced glutathione. B and C, total lysates of cells expressing His-tagged syndecan-2 were passed over a column of glutathione-agarose bound to GST (B) or GST-syndecan-2 (C), washed three times, and eluted. Each fraction was analyzed by SDS-PAGE followed by Coomassie Blue staining or Western blotting using the indicated antibodies. D and E, total lysates of cells expressing His-tagged syndecan-4 (D) or Histagged syndecan-2 (E) were passed over a column of glutathione-agarose bound to GST-syndecan-2 (D) or GST-syndecan-4 (E), washed three times, and eluted. Each fraction was analyzed by SDS-PAGE followed by Western blotting using the indicated antibodies.  SDS-PAGE. Half the gel was stained with Coomassie Blue to measure the amounts of homomeric (Fig. 5A, left, first and second lanes) and heteromeric (Fig. 5A, left, third lane) syndecans. Proteins in the other gel half were transferred to PVDF membranes (Fig. 5A, right); membranes were then incubated with B16F10 melanoma cell lysates, and interacting proteins were analyzed by Western blotting. Only homodimeric syndecan-4 mutants containing the cytoplasmic domain bound to PKC␣ (Fig. 5A, right); syndecan-2 homodimers and heterodimers of syndecan-2 and -4 did not. The increase in heterodimerization induced a significant decrease in syndecan-4 homodimers and led to a reduction in syndecan-4-PKC␣ interaction (Fig. 5B) and diminished PKC␣ activation (Fig. 5C, top), indicating an inhibitory effect of hetero-oligomerization on syndecan-4 function. Consistent with this, increased heterodimerization reduced the interaction of syndecan-4 with ␣-actinin, which is known to directly interact with the cytoplasmic domain of syndecan-4 (Fig. 5D).
Previous studies have shown that dimerization of PDGF receptor subsequently activates mitogen-activated protein kinase (MAPK) (16, 17). Moreover, a chimeric protein contain-ing the transmembrane domain of syndecan fused to the extracellular and cytoplasmic domain of ␤-PDGF receptor is sufficient to induced MAPK activation through chimera oligomerization (5). Accordingly, we explored the potential effect of transmembrane domain-mediated hetero-oligomerization on chimera-induced MAPK activation. HEK293T cells were transiently co-transfected with the chimeras PE2TPC and syndecan mutants containing extracellular and transmembrane domains of syndecan-4 (4ET), and chimera-induced MAPK activity was analyzed by Western blotting with an anti-phospho-ERK antibody. Consistent with previous data (5), phosphorylation of ERK was increased in chimera-transfected cells compared with that in vector-transfected cells (Fig. 5E). Co-transfection of the chimera with greater amounts of syndecan mutants caused a reduction in MAPK activity, implying the possibility of heterooligomer formation through the syndecan transmembrane domain and subsequent regulation of MAPK activation. Taken together, these results suggest that disruption of homo-oligomerization inhibits the function of individual syndecans.
Because syndecan-2 regulates tumorigenic activity in an oligomerization-dependent manner ( Fig. 7A; Refs. 22-24), we investigated whether hetero-oligomerization of syndecans might affect syndecan-2-mediated signal transduction involved in the regulation of tumorigenic activity. Co-localization of syndecan-2 and -4 was observed in HT29 colon adenocarcinoma cells transiently expressing syndecan-2 and -4 (Fig. 7B), suggesting that the two syndecans undergo hetero-molecular interactions in cancer cells. Similar to the observed effects on syndecan-4 signaling, our results showed that induction of hetero-oligomerization through co-overexpression of syndecan-2 and -4 significantly reduced the membrane localization of Tiam1 (T-cell lymphoma invasion and metastasis 1) (Fig. 7C) and activation of Rac (Fig. 7D), crucial regulators in syndecan-2-mediated signal transduction in colon cancer cells, compared with cells transfected with syndecan-2. Consistent with this, both syndecan-2-mediated cell migration and anchorage-independent growth were reduced in HT29 cells co-expressing syndecan-2 and -4 compared with those expressing syndecan-2 alone (Fig. 7, E-G), suggesting that hetero-oligomer formation inhibits syndecan-2 functions in colon cancer. Taken together, our findings indicate that hetero-oligomerization negatively regulates fundamental syndecan-mediated cellular processes.

Discussion
Syndecan core proteins are known to have the propensity to form non-covalently linked homodimers through interactions of a strictly conserved GXXXG motif in the transmembrane domain (4,5,25), but the potential of different syndecans to interact to form heterodimers has not been studied. Our investigation of the detailed interactions of syndecans reported here revealed that in addition to homodimer formation, all exogenously expressed syndecan-2 and -4 protein variants containing the transmembrane domain showed strong SDS-resistant heterodimer formation (Fig. 1). Consistent with this, endogenous syndecan-2 interacted with syndecan-4 ( Fig. 4) notwithstanding potential repulsive forces contributed by heparan sulfate chains and steric hindrance by the large syndecan extracellular domains, both of which would tend to prevent transmembrane domain association. We also found that syndecan-1, which has a low tendency to self-associate (11), showed SDS-resistant heterodimer formation with syndecan-2 and -4. Therefore, syndecan transmembrane domains appear to be involved in both homomeric interactions of a single syndecan type and heteromeric interactions between syndecan paralogs.
At least one syndecan is expressed in virtually all normal cells, but the expression pattern is distinct in each cell type. For example, epithelial cells express syndecan-1 but not syndecan-2. Thus, opportunities for formation of syndecan heterooligomers may be rare in normal cells; a notable exception is vascular smooth muscle cells, which express all four syndecan family members (26). However, under pathological conditions, syndecan expression is commonly altered. During carcinogenesis, cancer cells may undergo "receptor exchange," in which anti-tumorigenic receptors are down-regulated with the concomitant up-regulation of pro-tumorigenic receptors. For instance, the expression of syndecan-2, which is not expressed in normal epithelial cells, is increased during colorectal carcinogenesis (22). Altered syndecan-2 expression not only indicates that receptor exchange is sufficient for induction of colon tumorigenesis, it also suggests that the altered expression of syndecan-2 subsequently increases the likelihood of syndecan hetero-oligomerization. Syndecan hetero-oligomerization creates a new paradigm for integrating regulatory mechanism of syndecans in vivo. For example, syndecan-2 function has been studied independently from that of syndecan-4. However, hetero-oligomer formation between syndecan-2 and -4 raises the possibility of three cooperative regulatory mechanisms with FIGURE 6. Hetero-oligomerization regulates syndecan-4-mediated functions. A, scheme showing syndecan-4-mediated molecular mechanisms involved in adhesion signaling. B, cell-substrate adhesiveness was quantified using an inverted centrifugal detachment assay as described under "Experimental Procedures." The results are expressed as the mean percentages of retained cells per field Ϯ S.E. from three independent experiments. *, p Ͻ 0.01. C, REFs were co-transfected with the indicated cDNAs, and the amount of PKC␣ in the membrane fraction was determined by immunoblotting with the indicated antibodies. An antibody against integrin ␤1 (␣-Int␤1) was used as a loading control. D, REFs were co-transfected with indicated cDNAs. Cell lysates were subsequently prepared and incubated with purified GST-Rhotekin-Rho binding domain proteins for 2 h. Bound proteins were collected, and GTP-bound RhoA was detected by immunoblotting with anti-RhoA antibody. E, transfected cells were stained with anti-paxillin antibody (top). Bottom, percentages of cells showing focal adhesions (mean Ϯ S.E.; vector transfectants set as 100%). The data shown are representative of three independent experiments. *, p Ͻ 0.01. F, REFs co-transfected with syndecan-4 and either vector or syndecan-2 were plated on E-plates pre-coated with fibronectin (FN). After the indicated times cell spreading was monitored and plotted using the xCELLigence system. Representative results from three independent experiments are shown.
in cells: those involving homo-oligomers of syndecan-2, those involving homo-oligomers of syndecan-4, and those involving hetero-oligomers of syndecan-2 and -4. We hypothesized that syndecan hetero-oligomerization might reorganize the formation of functional complexes depending on the paralogs expressed and their relative abundance. Because transmembrane domain-induced homodimerization is important for the regulation of syndecan functions, the transition to heterodimers, which would necessarily constrain the formation of homo-oligomers, would be expected to disrupt syndecan functions. Consistent with this, our results revealed that heterooligomer formation between syndecan-2 and -4 inhibited both syndecan-4-mediated cell adhesion functions (Fig. 6) and syndecan-2-mediated tumorigenic activity (Fig. 7). These data suggest that increased hetero-oligomerization negatively regulates syndecan-mediated cellular signaling through restriction of homo-oligomerization. The oligomeric status of the syndecan-4 cytoplasmic domain is particularly important for syndecan-4 functions, including interactions with effector molecules. Hetero-oligomerization prevents homo-association of syndecan-4 cytoplasmic domains, thereby negatively regulating syndecan-4-mediated functions. These hetero-interactions may also contribute to novel signaling specificities because, in hetero-oligomers, the identity and geometry of ligand pairs differ from that of homo-oligomers. Distinct sets of syndecans formed through hetero-oligomerization could be regulated by different ligands, resulting in differential coupling to signaling effectors and biological responses. Dissecting the cooperative signal transduction pathways of syndecan homo-and heterooligomers would be an interesting avenue for future research.
Transmembrane domain interactions have been shown to play a functional role in other receptors, including growth factor receptor signal transduction. Binding of ligands to the extracellular domain of the epidermal growth factor receptor (EGFR or Erb) induces the formation of receptor homo-or heterodimers and subsequent activation of the intrinsic tyrosine kinase domain (27). A hierarchy of homomeric and heteromeric association has been identified for ErbB transmembrane domain peptides (28). These interactions may contribute to unique signaling specificities because kinase/substrate pairs in a trans-phosphorylation reaction within a heterodimer differ from those of a homodimer. In NIH3T3 cells ectopically expressing EGFR and ErbB3, neuregulin stimulates anchorageindependent growth, whereas EGF does not despite the fact that EGF stimulates higher levels of receptor phosphorylation than does neuregulin (29). Similarly, EGFR is activated by HB-EGF in a breast tumor cell line, leading to tyrosine phosphorylation of the downstream signaling effector Cbl and subsequent Cbl-EGFR interactions, whereas in EGFR-expressing cells in which EGFR is activated by neuregulin through transmodulation by ErbB3 or ErbB4, Cbl is not tyrosine phosphorylated and does not complex with the EGFR (30). These data suggest that heterodimerization mediates phosphorylation of distinct sets of receptor tyrosine residues in response to differ- C. HT29 cells were co-transfected with the indicated cDNAs, and the amount of Tiam1 in the membrane fraction was determined by immunoblotting with anti-Tiam1 antibody. An antibody against integrin ␤1 (␣Int␤1) was used as a loading control. D, purified GST-PAK-PBD protein was incubated with lysates from transfected cells for 2 h. Bound proteins were collected, and GTP-bound Rac was detected by immunoblotting with anti-Rac1 antibody. E, HT29 cells were transfected with the indicated cDNAs, and Transwell migration assays were performed as described under "Experimental Procedures." The data shown are representative of three independent experiments. *, p Ͻ 0.01. F, transfected cells were seeded in duplicate in the upper chamber of CIM-plates. The lower chambers were filled with medium containing 10% FBS and fibroblast growth factor-2 (10 g/ml), and migration was monitored and plotted using the xCELLigence system. Representative results from three independent experiments are shown. G, HT29 cells were seeded in soft agar and allowed to grow for 21 days, and the number of viable colonies was counted. Representative results from three independent experiments are shown. *, p Ͻ 0.01. ent ligands, creating an extracellular regulatory mechanism. Therefore, hetero-oligomerization through transmembrane domains might potentiate the diversification of biological functions.
This work clearly showed that syndecans form transmembrane domain-mediated heterodimers, and this dimerization regulates syndecan-mediated cellular functions. However, the effect of hetero-oligomerization might differ in various cells and tissues, perhaps due to tissue-specific differences in expression patterns and syndecan regulatory mechanisms. Further studies will be required to clarify the roles of hetero-oligomerization in specific syndecan regulatory mechanisms.